Tianshou: Basic API Usage

This tutorial is a simple example of how to use Tianshou with a PettingZoo environment.

It demonstrates a game betwenen two random policy agents in the rock-paper-scissors environment.

Environment Setup

To follow this tutorial, you will need to install the dependencies shown below. It is recommended to use a newly-created virtual environment to avoid dependency conflicts.

pettingzoo[classic]==1.23.0
packaging==21.3
tianshou==0.5.0

Code

The following code should run without any issues. The comments are designed to help you understand how to use PettingZoo with Tianshou. If you have any questions, please feel free to ask in the Discord server.

"""This is a minimal example to show how to use Tianshou with a PettingZoo environment. No training of agents is done here.

Author: Will (https://github.com/WillDudley)

Python version used: 3.8.10

Requirements:
pettingzoo == 1.22.0
git+https://github.com/thu-ml/tianshou
"""

from tianshou.data import Collector
from tianshou.env import DummyVectorEnv, PettingZooEnv
from tianshou.policy import MultiAgentPolicyManager, RandomPolicy

from pettingzoo.classic import rps_v2

if __name__ == "__main__":
    # Step 1: Load the PettingZoo environment
    env = rps_v2.env(render_mode="human")

    # Step 2: Wrap the environment for Tianshou interfacing
    env = PettingZooEnv(env)

    # Step 3: Define policies for each agent
    policies = MultiAgentPolicyManager([RandomPolicy(), RandomPolicy()], env)

    # Step 4: Convert the env to vector format
    env = DummyVectorEnv([lambda: env])

    # Step 5: Construct the Collector, which interfaces the policies with the vectorised environment
    collector = Collector(policies, env)

    # Step 6: Execute the environment with the agents playing for 1 episode, and render a frame every 0.1 seconds
    result = collector.collect(n_episode=1, render=0.1)